From: SB

To: Editor

 

Just a small question about internal data representation, a matter that doesn't seem to be considered much by most database folk, as they generally only worry about the input and output they see, rather than what the computer may be up to in the background.

 

Suppose I want to store a list of words, such as the colors of the rainbow for example, as a single column in a table, and use that column as the primary key. Will the computer access the list as a set of characters, employing all the various overheads of manipulating strings this would require, or does the computer access an integer list created internally as the key field?

 

I have always thought it best to add a second column of a numerical integer key to represent each entry, as an efficient means of data access. However, I have been aware that this introduces all sorts of problems, such as when inserting and deleting records, as well as the additional storage space and processing time. Now I am not sure that this is such a brilliant idea!

 

 

From: Fabian Pascal

To: SB

 

I am in agreement with Chris Date’s reply that I paraphrase:

 

1.      The logically correct answer is that the model and implementation should be kept separate at all times;

 

2.      The pragmatic answer is that with current products you sometimes are forced to violate that principle, which is always costly.

 

3.       The new implementation technology that permits to develop truly RDBMSs would obviate the need for that, but whether the industry will deploy it is a tossup.

 

 

Posted 11/08/02

 

 

 

[ABOUT] [QUOTES] [LINKS]